Studies from our laboratory as well as others over the past decade have confirmed the important participation of the amygdala in social perception, such as processing faces, and generated hypotheses about its role in mental illness, such as autism. Yet the precise role that the amygdala might play in mediating an endophenotype for social dysfunction remains poorly understood. This revised application plans a systematic and in-depth investigation of this topic in three specific aims. Aim 1 will investigate eye movements and emotional response during real social interactions (over live video and during face-to-face interactions). This will quantify the amygdala's contribution to realistic social interactions, and the impairments arising from autism. Aim 2 will investigate in more detail processing of social scenes containing faces, and tease apart possible aversion to faces from lack of attraction to them. Aim 3 will use the bubbles method to investigate which features in faces drive discrimination at the behavioral, emotional, and neural level. These experiments will be complemented by detailed questionnaires and interviews to assess real-life social functioning, and to explore individual differences. The aims will be carried out in neurological subjects with focal lesions of the amygdala, in high-functioning subjects with autism, in parents of autistic children who have the Broad Autism Phenotype, as well as in neurotypical controls. Contrasts between these groups will test the overall hypothesis that amygdala dysfunction contributes to the social impairments seen in autism as well as in first-degree relatives of people with autism. An endophenotype for impaired amygdala function is likely to contribute to several psychiatric illnesses in which the amygdala is thought to be dysfunctional, including depression, anxiety disorders, schizophrenia and autism. It is also likely to contribute to individual differences in the healthy population on dimensions that predispose to these psychiatric diseases, especially in those who share substantial genetic variance with these diseases (such as first-degree relatives). Our stimuli and methods will be made available to researchers and clinicians studying mental illness, and will inform diagnosis as well as provide a basis for designing future interventions.